L.-Shashua -Bar.pdf

15
Ž . Energy and Buildings 31 2000 221–235 www.elsevier.comrlocaterenbuild Vegetation as a climatic component in the design of an urban street An empirical model for predicting the cooling effect of urban green areas with trees L. Shashua-Bar a , M.E. Hoffman a,b, ) a Faculty of Architecture and Town Planning, Technion-Israel Institute of Technology, Haifa 32000, Israel b National Building Research Institute, Technion-Israel Institute of Technology, Haifa 32000, Israel Received 23 September 1998; accepted 14 May 1999 Abstract The cooling effect of small urban green wooded sites of various geometric configurations in summer is the object of this study. It was studied experimentally at 11 different wooded sites in the Tel-Aviv urban complex during the period July–August 1996. An empirical model is developed in this study for predicting the cooling effect inside the wooded sites. The model is based on the statistical analysis carried out on 714 experimental observations gathered each hour from the 11 sites on calm days, when urban climate is expressed. Two factors were found to explain over 70% of the air temperature variance inside the studied green site, namely, the partial shaded area under the tree canopy and the air temperature of the non-wooded surroundings adjoining the site. The specific cooling effect of the site due to its geometry and tree characteristics, besides the shading, was found to be relatively small, about 0.5 K, out of an average cooling of about 3 K at noon. The cooling effect of the green wooded areas on their immediate surroundings at noon was also analyzed. The findings corroborate earlier studies that the range is noticeable. At small green sites, the cooling effect estimated in this study is perceivable up to about 100 m in the streets branching out from the site. The empirical findings in this study permit development of tools for incorporating the climatic effects of green areas in the urban design. Some policy measures are proposed accordingly, for alleviating the ‘‘heat island’’ effect in the urban environment. q 2000 Published by Elsevier Science S.A. All rights reserved. Keywords: Urban climate; Trees; Wooded site cooling; Surrounding cooling effect; In situ experimental study; Empirical thermal quantification models 1. Introduction As urbanization progresses, the ‘‘heat island’’ problem is aggravated mainly because of the reduced density of the wx green vegetation in the urban environment 1 . Additions to public green areas usually lag behind the urban develop- ment. Private green areas in the courtyards of apartment houses are also declining, due to conversion for parking purposes. The reduction in the green area densities has an adverse w x effect on local air temperature. Rosenfeld et al. 2,3 illustrate the case for downtown Los Angeles over the period 1882–1984. With increasing irrigation and or- chards, the city of Los Angeles cooled by 2 K until the ) Corresponding author. National Building Research Institute, Tech- nion-Israel Institute of Technology, Haifa 32000, Israel. Tel.: q 972-4- 829-2285; fax: q 972-4-832-4534; E-mail: [email protected] 1930s. Since then, as asphalt replaced trees, the city warmed by 3 K. The phenomenon is universally typical. In macro, the control measures suggested are mainly vegeta- wx tion and high-albedo roofs and streets. Rosenfeld et al. 2 study in alleviating the heat-island problem, believe in a three-pronged strategies beyond microclimate below trees: Ž. Ž. Ž. a cool roofs, b cool pavements, and c vegetation for evapotranspiration. Vegetation surfaces show lower radiative temperatures than other inanimate ones of the same colour. The differ- wx ence in maximum temperature may exceed 20 K 4 . In the case of large green areas such as parks, vegetation affects the air temperature above it and thus improves the thermal wx environment of the urban area. Jauregui 5 found that in Ž . Chapultepec Park 500 ha in Mexico City, the effect of the park on air temperature is noticeable at a radius of 2 km, about the same as its width. In a new paper on Tama wx New Town’s Central Park in Japan by Ca et al. 6 found Ž . that the influence area of the park about 35 ha can extend 0378-7788r00r$ - see front matter q 2000 Published by Elsevier Science S.A. All rights reserved. Ž . PII: S0378-7788 99 00018-3

Transcript of L.-Shashua -Bar.pdf

Ž .Energy and Buildings 31 2000 221–235www.elsevier.comrlocaterenbuild

Vegetation as a climatic component in the design of an urban streetAn empirical model for predicting the cooling effect of urban green

areas with trees

L. Shashua-Bar a, M.E. Hoffman a,b,)

a Faculty of Architecture and Town Planning, Technion-Israel Institute of Technology, Haifa 32000, Israelb National Building Research Institute, Technion-Israel Institute of Technology, Haifa 32000, Israel

Received 23 September 1998; accepted 14 May 1999

Abstract

The cooling effect of small urban green wooded sites of various geometric configurations in summer is the object of this study. It wasstudied experimentally at 11 different wooded sites in the Tel-Aviv urban complex during the period July–August 1996. An empiricalmodel is developed in this study for predicting the cooling effect inside the wooded sites. The model is based on the statistical analysiscarried out on 714 experimental observations gathered each hour from the 11 sites on calm days, when urban climate is expressed. Twofactors were found to explain over 70% of the air temperature variance inside the studied green site, namely, the partial shaded area underthe tree canopy and the air temperature of the non-wooded surroundings adjoining the site. The specific cooling effect of the site due to itsgeometry and tree characteristics, besides the shading, was found to be relatively small, about 0.5 K, out of an average cooling of about 3K at noon. The cooling effect of the green wooded areas on their immediate surroundings at noon was also analyzed. The findingscorroborate earlier studies that the range is noticeable. At small green sites, the cooling effect estimated in this study is perceivable up toabout 100 m in the streets branching out from the site. The empirical findings in this study permit development of tools for incorporatingthe climatic effects of green areas in the urban design. Some policy measures are proposed accordingly, for alleviating the ‘‘heat island’’effect in the urban environment. q 2000 Published by Elsevier Science S.A. All rights reserved.

Keywords: Urban climate; Trees; Wooded site cooling; Surrounding cooling effect; In situ experimental study; Empirical thermal quantification models

1. Introduction

As urbanization progresses, the ‘‘heat island’’ problemis aggravated mainly because of the reduced density of the

w xgreen vegetation in the urban environment 1 . Additions topublic green areas usually lag behind the urban develop-ment. Private green areas in the courtyards of apartmenthouses are also declining, due to conversion for parkingpurposes.

The reduction in the green area densities has an adversew xeffect on local air temperature. Rosenfeld et al. 2,3

illustrate the case for downtown Los Angeles over theperiod 1882–1984. With increasing irrigation and or-chards, the city of Los Angeles cooled by 2 K until the

) Corresponding author. National Building Research Institute, Tech-nion-Israel Institute of Technology, Haifa 32000, Israel. Tel.: q972-4-829-2285; fax: q972-4-832-4534; E-mail:[email protected]

1930s. Since then, as asphalt replaced trees, the citywarmed by 3 K. The phenomenon is universally typical. Inmacro, the control measures suggested are mainly vegeta-

w xtion and high-albedo roofs and streets. Rosenfeld et al. 2study in alleviating the heat-island problem, believe in athree-pronged strategies beyond microclimate below trees:Ž . Ž . Ž .a cool roofs, b cool pavements, and c vegetation forevapotranspiration.

Vegetation surfaces show lower radiative temperaturesthan other inanimate ones of the same colour. The differ-

w xence in maximum temperature may exceed 20 K 4 . In thecase of large green areas such as parks, vegetation affectsthe air temperature above it and thus improves the thermal

w xenvironment of the urban area. Jauregui 5 found that inŽ .Chapultepec Park 500 ha in Mexico City, the effect of

the park on air temperature is noticeable at a radius of 2km, about the same as its width. In a new paper on Tama

w xNew Town’s Central Park in Japan by Ca et al. 6 foundŽ .that the influence area of the park about 35 ha can extend

0378-7788r00r$ - see front matter q 2000 Published by Elsevier Science S.A. All rights reserved.Ž .PII: S0378-7788 99 00018-3

( )L. Shashua-Bar, M.E. HoffmanrEnergy and Buildings 31 2000 221–235222

to a distance of 1 km in the northwest direction when thewind is very strong.

In micro, the effect of vegetation on the thermal envi-ronment of its surroundings area is rather small but still

w xsignificant. In Israel, Givoni 7 found that the coolingŽ .effect of Haifa’s Biniamin Park 0.5 ha is noticeable 20 to

w x150 m outside it. In Japan, Hunjo and Takakura 8 , usinga numerical model, showed a range of 200 m in thedirection of the wind, the width of the green area being300 to 700 m. The results of their simulations indicate thatthe range of the effect is a function of the green area scaleand the intervals between the green areas. They suggestthat smaller green areas with sufficient intervals are prefer-able for effective cooling of the surroundings to lumpedlarger green areas.

The cooling effect in small areas is obtained mainlyw xthrough shading 9,10 . Other factors that inhibit penetra-

tion of solar radiation, besides shading, may also play arole in determining the cooling effect of a green site. Thegeometric configuration may also affect temperature varia-tions, as was found to be the case in non-wooded building

w x 1structures 11,12 .In the present project, to allow for geometric variations,

the empirical study covered a set of urban green habitats ofdifferent sizes, shapes and built-up morphology. Measure-ments of air temperature, humidity, wind velocity, solarradiation penetration and surface radiant temperature 2

were carried out at the sites during the summer of 1996.The object of this empirical study was to determine the

factors affecting the microclimate inside the green site andits influence on the surrounding areas. On the basis of thestatistical findings, an empirical model was developed forpredicting the maximum cooling effect inside the site andits range outside the site. The model may be useful incost–benefit analysis in designing a green area. 3

2. Methodology

Statistical analysis of the collected experimental data onair temperature and humidity inside and outside the siteswas carried out with the aid of linear regression models.The relationships to be explained are the cooling andhumidity effects of the site on its own microclimate and onits immediate surroundings.

1 The geometric configuration of a ‘‘canyon’’-form street in Swaid andŽ .Hoffman’s cluster thermal time constant CTTC model is represented by

the heightrwidth ratio and by the sky view factor.2 Measurements of solar radiation penetration and surface temperature

at the studied sites are being considered for use in further development ofan analytical urban climate CTTC model, previously developed in former

Ž w x.works see Refs. 8–10 .3 Following the empirical model, a more general analytical model is

Ž w x.being developed based on the CTTC model see Refs. 8–10 and will bediscussed in a separate paper.

Comparison of these effects among different sites isproblematic. From the publications cited above it is knownthat the air temperature inside the site depends on the

Ž .shading intensity partial shaded area , on the thermalproperties of the soil, and on the air temperature of itsimmediate background. The latter varies among differenturban sites due to factors affecting the air temperature suchas vegetation, built-up geometry, topography, traffic den-sity and other anthropogenic heat-release factors. 4 Com-parison of the sites’ air temperatures to that of a singleoutside reference point such as a nearby meteorologicalstation would lead to wrong conclusions in our case, evenwhen the comparison is done for days of measurements ata single site. To overcome this problem, the cooling effectwas considered in this study as the difference between theair temperature measured at the site and that at the corre-sponding ‘‘reference point’’, chosen so as to comply withthe following two criteria:Ž . Ža The reference point is close to the site 50 to 100 m

.from it .Ž .b The reference point is treeless and receives sunshinemost of the day.

Thus for each site a ‘‘reference point’’ is selected. Itrepresents the site background without vegetation effects.

The cooling effect so defined is still to some extent afunction of the surrounding background temperature. Thisbackground factor will be estimated and incorporated inthe empirical model. We note, parenthetically, that whenthe background air temperature variation among the sites is

Ž .relatively small say 1 K to 2 K , the suggested measuresimplifies the comparison without further ado.

The humidity effect of the site is evaluated in the sameŽway, as the difference between the humidity absolute or

.relative of the site and that of its reference.

3. Sites and observations

The present study was conducted during the summerŽ .July–August of 1996 on 11 urban green areas with trees,chosen so as to represent a variety of typical areas such assmall gardens and courtyards, avenues with and withouttraffic and ‘‘canyon’’ streets with trees. The sites werelocated in the so-called Dan complex, consisting of Tel-Aviv proper and the adjoining cities of Givatayim and

Ž .Ramat-Gan Fig. 1 . This region is characterized by analmost uniform topography and small climatic variationsfrom day to day during the summer season. Consequentlyone would expect small variations in the climatic factorsamong the sites’ backgrounds.

4 In a private letter, Rosenfeld mentions that in Los Angeles in summeranthropogenic heat-release factors are a few percent effect. This may alsobe the case in the Dan complex studied in this paper.

( )L. Shashua-Bar, M.E. HoffmanrEnergy and Buildings 31 2000 221–235 223

Fig. 1. The observation sites.

At each site, several observation points, spaced at about20 m, were chosen inside it over its length and severalpoints outside. Temperature measurements were taken withwind velocity not exceeding 0.5 mrs in the sites. On thedays of measurement, the Standard Meteorological windvelocity was between 1 and 2 mrs during the day andcalm, less than 0.5 mrs, during the evenings and nights.No measurements were taken on windy days.

Ž .Dry and wet bulb temperatures DBT and WBT , weremeasured at 0600, 0900, 1500, 1800 and 2400 h, approxi-mately at the height of 1.80 m. Emphasis was on the 1500Ž .h 1410 solar time data, representing the maximum daily

temperature. Besides the climatic variables, the partialshaded area around each observation point in the site wasalso determined at 1500 h.

Temperatures were measured with a DB–WB Slinghygrometer. Parallel readings were taken for comparison

Ž .with the aid of a digital thermometer DT with its sensorshielded in an aluminum foil cylinder 3 cm in diameter 5cm in length, open on both sides and painted white on theoutside. Measurements with the DBT and DT devicesshowed almost insignificant differences throughout. Tem-perature and humidity were measured in the shade. Winddirection and velocity were measured in parallel with the

Žaid of a cup anemometer. Solar radiation intensity direct.and transmitted through the tree canopy was measured

with a Kipp and Zonen solar pyranometer. Surface temper-Žatures ground, tree trunks, leaves above and below the. Ž .canopy were measured with an infrared IR thermometer

calibrated against the DBT thermometer bulb. The IRthermopile detector spectral response in the range of 7–18mm. 2

Table 1 lists the number of observation points insideŽ .each site 100 points altogether and the corresponding

Table 1Site data and measurement plan

Ž .Site Width of Length of Number of Number of Dates of measurements 1996a b cŽ . Ž .site m site m observation points observations

Ž .1 Hayeled Avenue 22 200 14 140 July 2, 5, 10, 12, August 18Ž .2 Meltz Garden 35 112 10 80 August 2, 4, 6, 8Ž .3 Emanuel Avenue 30 115 9 72 August 2, 4, 6, 8Ž .4 Rothschild Avenue 45 245 18 108 July 25, 28, 29Ž .5 Hen Avenue 35 250 17 102 July 25, 28, 29Ž .6 Borochov Square 60 60 4 24 July 22, 23, 31Ž .7 Courtyard A 15 30 4 16 August 11, 13Ž .8 Courtyard B 20 25 4 16 August 11, 13Ž .9 Aharon Garden 40 25 4 16 August 26, 27Ž .10 K.K.L. Street 30 90 5 30 July 22, 23, 31Ž .11 Herzl Street 20 220 11 110 July 2, 5, 10, 12, August 18Total – – 100 714 –

a Where readings were taken.bSpacing 20 m.c Not included — observation points outside the sites.

( )L. Shashua-Bar, M.E. HoffmanrEnergy and Buildings 31 2000 221–235224

Fig. 2. Site maps.

( )L. Shashua-Bar, M.E. HoffmanrEnergy and Buildings 31 2000 221–235 225

total number of observations over the period of analysisŽ .714 each hour . Not listed in the table are about 45 points

Ž .outside the sites see further discussion in Section 4 . SiteŽ .width ranged between 15 m courtyards A and 60 m at

Borochov Square, the intersection of four streets. Thebuildings bordering the sites are 12 to 15 m high.

The sites include two ‘‘canyon’’ streets with trees alongŽ .the sidewalks Herzl and K.K.L. . The two avenues

Ž .Rothschild and Hen also have trees along a 12-metermedian strip. These avenues and Herzl Street carry heavytraffic. The other sites are gardens and avenues closed tovehicle traffic.

The site maps are given in Fig. 2, with reference pointsdenoted by R. All site axes are oriented close to North–South. All sites are planted with Ficus trees about 50 to 70years of age, except for Meltz Garden and Herzl Street,

which are planted with a variety of trees with Poincianapredominating.

Some preliminary results of the thermal and humidityeffects inside the sites are summarized in Figs. 3 and 4 andin Tables 2 and 3.

The pattern of the cooling effect along Hayeled Avenuesite is shown in Fig. 3, where point 1 is the referencepoint, and point 2 is about 40 m outside the entrance to thesite, bordering the street, under a large Ficus tree. Saito et

w x w xal. 1 and Rosenfeld et al. 2 found that even a single treecan affect the air temperature of the immediate surround-ing area. In Fig. 3, the cooling effect at point 2 is about 1K at 1500 h, compared to the maximum cooling effect of2.5 K at point 8 inside the site.

Similar cooling effect patterns were found in all otherŽ .10 sites Fig. 5 . The weakest cooling effect is found near

w x Ž .Fig. 3. The daily cooling effects along the Hayeled avenue site K averages for the days of measurement .

( )L. Shashua-Bar, M.E. HoffmanrEnergy and Buildings 31 2000 221–235226

w x w xFig. 4. Comparison between cooling effects K and relative humidity differences percentage points measured along the Hayeled avenue; time 1500 h.

the entrance, and the strongest one inside the site aroundthe midsection.

Further statistical analysis of the factors determiningthese effects is given in Section 3.

The maximum cooling effects obtained inside the 11sites are summarized in Table 2. Each figure is the averagefor the days of measurement at the coolest point insideeach site, usually at its midsection. The average effect at1500 h is about 3 K, ranging from 1.3 K for Herzl Street,

Ž .to about 4 K for the small Aharon garden 0.15 ha . NoŽ .significant cooling is found before sunrise 0600 h , or at

Ž .midnight 2400 h . The average levels outside the sites forthe whole surveyed season were between 24.58C at sunriseand 32.78C at noon.

The partial water vapour pressure for all observationswas calculated from the DB and WB temperatures, with noconsistent significant differences found between the sitesand their respective reference points. This may be due tothe fact that except for the Melts garden and the twocourtyards, these sites are not irrigated. The absence ofevapotranspiration change within the site does not mean noevapotranspiration. It indicates a low rate of evapotranspi-

w xration which as Oke 10 observed ‘‘probably occurs fromthe top of the trees and does not mix throughout thevolume’’ under the tree canopy. The effect of evapotran-spiration, in our case, was expressed in cooling the canopyleaves. The leaves surface temperature, at noon, relative to

Ž .the reference air temperature At Hayeled Avenue wasy3.1 K under the canopy and y1.1 K above the canopy.Without wind and without the evapotranspiration effectsay in dead leaves the surface irradiant temperature of theleaves above the canopy would have been much higher

w xthan the air temperature, probably exceeding 20 K 4 .With regards to relative humidity, significant differ-

ences were found: the air temperature inside the site islower than at the reference point, the saturation vapourpressure is also low and the corresponding relative humid-ity inside the site is consequently high. As the vapourpressure almost generally does not vary, the relative hu-midity differences along the length of the site follow a

pattern similar to the cooling effect, with the reverse signas expected. The pattern of the maximum humidity differ-ences and the maximum cooling effect at noon for theHayeled site are presented in Fig. 4. The maximum humid-ity difference at noon, at observation point 8, is about 6percentage points. Similar relative humidity patterns as inFig. 3 were obtained at all the other sites.

Table 3 shows the averages of the maximum relativehumidity effects at the various sites, calculated as percent-

Žage points the difference between the relative humidity.percentage inside the site and at its reference . The maxi-

mum effect at 1500 h was about 10 percentage points atRothschild Avenue, Aharon garden and courtyard B. Ta-bles 2 and 3 indicate that the average cooling at 1500 hinside the sites is about 2.8 K and the humidity effect 7.7percentage points. These effects are to be related to theaverage levels at the reference points. The average air

Žtemperature at 1500 h outside the sites was 32.78C 31.88C. Žto 33.78C and the average relative humidity 66.4% 63%.to 69.1% for the whole surveyed period.

The relationship between the cooling and the higherrelative humidity effects at 1500 h, as shown in Fig. 4, was

Table 2w x ŽDaily maximum cooling effects, K averages for the days of measure-

.ment inside the sites

Site 0600 h 0900 h 1500 h 1800 h 2400 h

Ž .1 Hayeled Avenue y0.10 y1.30 y2.50 y1.20 y0.20Ž .2 Meltz Garden y0.70 y2.30 y2.90 y2.40 y1.20Ž .3 Emanuel Avenue y0.40 y2.10 y3.30 y2.20 y0.70Ž .4 Rothschild Avenue – y1.40 y3.20 y1.80 –Ž .5 Hen Avenue – y1.40 y2.80 y1.70 –Ž .6 Borochov Square – y1.30 y2.90 y1.20 –Ž .7 Courtyard A – y1.50 y2.50 y2.30 –Ž .8 Courtyard B – y1.80 y3.40 y2.60 –Ž .9 Aharon Garden – y2.30 y4.00 y1.60 –Ž .10 K.K.L. Street – y1.00 y2.30 y1.10 –Ž .11 Herzl Street 0.30 0.20 y1.30 y0.50 0.20Average y0.20 y1.50 y2.80 y1.70 y0.50

–: Not measured.

( )L. Shashua-Bar, M.E. HoffmanrEnergy and Buildings 31 2000 221–235 227

Table 3w x ŽDaily relative humidity effect percentage points averages for the days.of measurement inside the site

Site 0600 h 0900 h 1500 h 1800 h 2400 h

Ž .1 Hayeled Avenue 1.0 5.9 6.0 2.8 1.1Ž .2 Meltz Garden 4.7 10.0 9.4 9.8 7.2Ž .3 Emanuel Avenue 1.2 8.6 9.9 7.0 4.2Ž .4 Rothschild Avenue – 3.8 10.4 6.4 –Ž .5 Hen Avenue – 3.3 9.4 5.2 –Ž .6 Borochov Square – 6.6 8.9 5.6 –Ž .7 Courtyard A – 6.1 5.2 10.8 –Ž .8 Courtyard B – 8.6 10.9 12.7 –Ž .9 Aharon Garden – 7.4 10.2 6.2 –Ž .10 K.K.L. Street – 4.9 3.4 3.6 –Ž .11 Herzl Street 0.9 y0.4 1.2 0.2 y1.3Average 1.9 5.9 7.7 6.4 2.8

approximated by a linear regression. The slope parameterfor all sites was found to be y3.2 percentage points foreach degree centigrade of cooling, as expected from psy-chometric equations. The average correlation coefficientwas y0.860 — highly significant, as expected.

4. Analysis of the cooling effect

The multiple linear regression method was applied inestimating the cooling effect inside the sites. The explana-tory variables considered are shading coverage, back-

Ž .ground reference air temperature, and the site specificeffect.

The shading coverage plays an important role in pre-dicting the air temperature inside a site in analytical mod-

Ž w x.els e.g., Refs. 7,10 . In a green area with trees, thecooling effect is determined by the amount of canopyshading. Different levels of shading will produce different

Ž .levels of the cooling effect: for example Fig. 3 thefluctuations of the cooling effect along Hayeled Avenueare due to non-uniform shading.

The shading coverage effect expresses the effect of theŽfactors governing the penetration of solar radiation per-

.meability . The canopy shading is determined inter alia bycanopy shape and depth and leaf area distribution, spacingof the trees, and growth factors such as cultivation andirrigation regime.

The second explanatory variable considered is the back-ground air temperature. The contribution of this variable isvery important for comparison of the measurements takenon different days at a particular site, as well as amongsites.

The third variable in the regression model represents thecooling effect, here the so-called ‘‘site specific effect’’. Itencompasses the effect of all unspecified variables govern-ing the site microclimate such as geometric configuration,tree characteristics and growth factors. The effect of theunspecified variables is represented by the constant termwhich is the intercept of the regression line.

The three explanatory variables were found to be inde-pendent. The correlation between shading coverage andbackground air temperature is zero. Consequently, theregression coefficients of the multiple linear model can beestimated separately by simple regressions. This procedurehas the advantage of simplicity but more important, itallows to judge whether the estimated coefficients differamong the sites.

4.1. The tree canopy shading effect

The shading effect at noon was estimated for each siteseparately from measurements along it, using linear regres-sions:

DT sa qb PSA 1Ž .ŽÕyr . , j ,Õ Õ 2 j ,Õ

where j,Õ is the observation point j at site Õ, r is themeasurement at the reference, DT sT yT isŽÕyr ., j,Õ j,Õ r ,Õ

the average cooling effect at point j of site Õ, and PSA j,Õ

is the partial shaded area around the jth observation pointof site Õ. The method of estimating PSA is described inFig. 6. In this work, at each site the observation pointswere chosen spaced at about 20 meters inside it over its

Ž .length. The area around the observation point X wasdivided into n usually 5 to 10 rectangular strips of equalwidths.

The shading coefficients b for the various investigated2

sites are shown in Table 4. The observations used in theseregressions were the averages for the measurement days atthe n observation points along each site. PSA has thej,Õ

same value for all measurement days and the same numberof days for all observation points. Thus, the coefficientestimated from the averages of the cooling effect yieldsexactly the same value as that derived from the dailyobservations.

All correlations are statistically highly significant. OnŽ 2the average, this factor accounts for about 70% r s

2 .0.83 =100 of the cooling effect variance for the studiedsites. The shading coefficient b was found to have about2

Ž .the same magnitude y3.23 at all sites, except at theŽ .Hayeled Avenue y2.15 . Thus complete shading cover-

age from the canopy is expected to have an averagecooling effect of 3.23 K at all the sites and of about 2.15 K

Žat Hayeled Avenue probably because this site was not.irrigated .

w xSaito et al. 1 found a similar effect between daytimeair temperature and green coverage ratio at the city ofKumamoto, Japan. The average effect of the green areasfound there was b sy1.7 as against to y3.2 found here.2

The constant term a calculated by the regression inÕ

Ž . ŽEq. 1 is the estimate of the ‘‘site specific effect’’ see.Section 4.4 . With b known, a can be obtained from:2 Õ

a sDT yb PSA 2Ž .Õ ŽÕyr . 2 Õ

where DT and PSA are the averages over all observa-ŽÕyr . Õ

tion points of site Õ, at 1500 h, for all the measurementdays.

( )L. Shashua-Bar, M.E. HoffmanrEnergy and Buildings 31 2000 221–235228

wx

Fig.

5.D

aily

cool

ing

effe

cts

alon

gth

esi

tes:

K.

( )L. Shashua-Bar, M.E. HoffmanrEnergy and Buildings 31 2000 221–235 229

Ž.

Fig.

5co

ntin

ued

.

( )L. Shashua-Bar, M.E. HoffmanrEnergy and Buildings 31 2000 221–235230

Ž .Fig. 6. Estimation of Partial Shaded Area PSA . Denoting by: dsaveragelength of each strip, equal to the site width, around the observation pointŽ .X . d saverage unshaded length of the ith strip. The partial unshadedi

Ž . Ž .area PUSA around the observation point X is: PUSAsS d rnd. Thei

partial shaded area PSA is: PSAs1y PUSA. Judgement was allowed inestimating the equivalent average length of sunny spots along a strip.

According to the results in the last column in Table 4,the average partial shaded area at 1500 h at all sites was

Ž .about 61% PSAs0.609 and contributed to about 80% ofthe total cooling effect of the sites.

The four sites: Borochov Square, Aharon Garden,K.K.L. and Herzl Streets, were excluded from the analysisbecause of the minor variation found in the PSA amongthe observation points along each of them. 5 In calculatinga for these four sites, b was taken equal to the averageÕ 2

Ž .for the sites in Table 4 y3.23 . a for the four sites wasÕ

Ž .calculated according to Eq. 2 and was found to be:Ø Borochov Square: a sy0.57Õ

Ø Aharon Garden: a sy0.56Õ

Ø K.K.L. Street: a sy0.02Õ

Ø Herzl Street: a sq1.03Õ

The first three values of a are of the same order ofÕ

magnitude as those for the other sites in Table 4 and havethe same sign. This fact confirms our assumption that b2

has about the same value for all studied sites. The a forÕ

Herzl Street is different, and probably reflects the heavytraffic in daytime.

4.2. The background effect

The background effect was assumed to have the samevalue for all sites. It was estimated by regressing thecooling effect at the site’s coolest spot as the dependentvariable with respect to the air temperature at the referenceas the explanatory variable:

DT sconstantqb T 3Ž .ŽÕyr . , i , j ,Õ 1 i ,r ,Õ

where point j is the coolest spot of site Õ in the ith day ofmeasurement, usually found around the middle of the site.This place was chosen for the analysis to emphasize thebackground effect.

5 In Herzl Street, the PSA level was 0.65 at nine observation pointsand 0.55 at other two points. In Aharon Garden, the PSA was 0.90 atthree observation points and 0.75 at a fourth. Such slight variation in theexplanatory variable is not sufficient for significant results in the regres-sion analysis.

The background effect was estimated separately at 0900,1500 and 1800 h. The results are given in Table 5. Thenumber of observations was 44 for each hour. All correla-tions are statistically highly significant.

The background coefficient b is seen to vary over the1Ž .day. It is small at noon b sy0.315 and is about the1

Ž . Žsame at 0900 h b sy0.515 and at 1800 h b s1 1.y0.636 . These findings clearly indicate that the back-

ground air temperature affects the level of cooling insidethe site. The higher the background air temperature, thestronger the cooling effect. For example, when the back-ground temperature rises by say 10 K, the cooling effect atnoon is enhanced by about 3.15 K. The difference in thecooling effect between any two sites Õ and Õ , due to the1 2

difference in their background air temperatures, equalsŽ .b T yT .1 r ,Õ r ,Õ1 2

4.3. The site specific effect

Ž .The site specific effect a , as estimated by Eq. 1 , isÕ

affected by the site’s average background air temperatureŽ Ž ..T in the same way as the cooling effect DT Eq. 3 . Tor ,Õ

compare the a s among the various sites, they are rescaledÕ

Ž .to relate to a single reference air temperature T whichÕ,r

is the average of all sites’ reference air temperatures, asfollows:

A sa yb T yT 4Ž .Ž .Õ Õ 1 r ,Õ r

where A is the rescaled specific effect.Õ

The rescaled site specific effect A 6 relative to theÕ

average background air temperature at 1500 h during theŽ .surveyed season T s32.78C is given in Table 6 and isr

Ž .on the average about y0.5 K except for the streets . Thesite specific effects are relatively small and all have the

Ž .correct negative sign except for courtyard A . On theaverage, the site specific effect, apart from shading, con-

Žtributes about 18% of the total cooling effect see last.column in the table . In contrast, the site specific effect of

ŽHerzl Street is positive about 0.75 K see Section 4.1.above . Accepting the assumption that the shading effect

Ž .of trees coefficient b at this site is the same as for the2Ž .others y3.23 , the positive effect of 0.75 K may reflect

the heavy traffic. Relative to the average site specificeffect value of about y0.5 K, the effect of the traffic inHerzl Street is about 1.2 K. This traffic effect is notnoticeable in the two avenues — Rothschild and Hen, norat Borochov Square. These three sites are much wider thanHerzl Street and their tree canopies are higher, so that thesuperior cooling effect may be due to the stronger ventila-

6 Ž .In calculating a relation 2 the shading coefficient used is b sÕ 2

y2.15 for the Hayeled Avenue and b sy3.23 for all the other sites.2

( )L. Shashua-Bar, M.E. HoffmanrEnergy and Buildings 31 2000 221–235 231

Table 4Ž . Ž .Results of regression analysis between cooling effect DT and partial shaded area PSA for the different sites; time: 1500 hŽÕyr .

Ž . wŽ . xSite DT K PSA n a b r b PSA 100 rŽÕyr . Õ Õ 2 2 Õ

Ž .DTŽÕyr .

Ž .1 Hayeled Avenue y1.919 0.650 14 y0.5260 y2.15 0.865 72.8Ž .2 Meltz Garden y2.248 0.460 10 y0.8181 y3.11 0.760 63.6Ž .3 Emanuel Avenue y2.470 0.633 9 y0.5230 y3.07 0.901 78.7Ž .4 Rothschild Avenue y2.143 0.622 18 y0.0696 y3.33 0.937 96.6Ž .5 Hen Avenue y2.375 0.635 17 y0.3078 y3.25 0.786 86.9Ž .7q8 Courtyard AqB y2.476 0.656 8 y0.2680 y3.37 0.830 89.3Average y2.272 0.609 76 – – 0.830 81.3

tion. In K.K.L. Street, the traffic is minimal and conse-quently the effect is negligible.

4.4. The cooling effect model

Combining the shading and background effects of Eqs.Ž . Ž .1 and 3 , we have:

DT sa qb T yT qb PSA 5Ž .Ž .ŽÕyr . , i , j ,Õ Õ 1 i ,r ,Õ r ,Õ 2 j ,Õ

where the subscript i denotes the measurement day.Ž . Ž .In Eq. 5 , the term b T yT represented the1 i,r ,Õ r ,Õ

effect of the site’s background air temperature deviation onday i on the cooling effect. Thus, the day-to-day coolingeffect variations, at a certain hour, are due merely tochanges in the background air temperatures. As this effect

Ž .is measured relative to the site’s average reference T ,r ,ÕŽ .the term T yT cancels out on the average.i,r ,Õ r ,Õ

The estimating model where the background effects arerelated to a single background air temperature T , is givenr

Ž .in Eq. 6 :

DT sA qb T yT qb PSA 6Ž .Ž .ŽÕyr . , i , j ,Õ Õ 1 i ,r ,Õ r 2 j ,Õ

Ž .Note that in Eq. 5 the average temperature at the sitereference T is used for estimating the cooling effectr ,Õ

Ž .while in Eq. 6 the average background air temperature Tr

is used, hence, the use of A instead of a .Õ Õ

Ž . Ž .Both Eqs. 5 and 6 yield the same estimate for thecooling effect. However, as the site’s average backgroundair temperature T is usually not known in advancer ,Õ

Ž .whereas T which is characteristic of the region is known,rŽ .the rescaled model in Eq. 6 is preferable, apart from the

fact that A is meaningful for comparison purposes amongÕ

the studied sites while a is not.Õ

Substituting the estimated parameter values for 1500 hŽ .according to Eq. 6 , the empirical cooling effect model for

the studied sites, related to a single background air temper-Ž .ature T s32.78C is given below in Eq. 7 :r

DT s10.3qA y0.315T qb PSA 7Ž .ŽÕyr . , i , j ,Õ Õ i ,r ,Õ 2 j ,Õ

where 10.3syb T sy0.315)32.78C, b sy2.15 for1 r 2

the Hayeled Avenue; b sy3.23 for the other sites.2

5. Thermal effects of green sites on their surroundingareas

The ‘‘surrounding area’’ of a site is defined in thisŽstudy as the area just outside the entrance to the site case

. Ž .A or a street crossing the site case B . These areas haveno green vegetation.

For each site, several outside points of measurementwere chosen at 20 m intervals. The air temperature mea-surements at these points were taken at the same hourbetween 1430 and 1530 h as at the other sites. The coolingeffect was calculated as the difference between the airtemperature at each observation point and that at thereference.

Table 5Regression results of the background effect

0900 h 1500 h 1800 h

Constant, a 13.19 7.32 17.57Slope, b y0.515 y0.315 y0.636Correlation, r y0.835 y0.547 y0.800

( )L. Shashua-Bar, M.E. HoffmanrEnergy and Buildings 31 2000 221–235232

Table 6The specific cooling effects of the sites; time: 1500 h

Ž . Ž . Ž . Ž .Site n DT K T 8C PSA A K A =100 rŽÕyr . r ,Õ Õ Õ Õ

Ž .DTŽÕyr .

Ž .1 Hayeled Avenue 14 y1.92 31.8 0.65 y0.81 42.2Ž .2 Meltz Garden 10 y2.25 33.2 0.46 y0.61 27.5Ž .3 Emanuel Avenue 9 y2.47 33.2 0.63 y0.28 11.3Ž .4 Rothschild Avenue 18 y2.14 32.3 0.62 y0.26 12.1Ž .5 Hen Avenue 17 y2.38 32.3 0.64 y0.44 18.5Ž .6 Borochov Square 4 y2.37 32.3 0.56 y0.68 25.3Ž .7 Courtyard A 4 y2.49 33.7 0.75 q0.25 y10.0Ž .8 Courtyard B 4 y3.28 33.7 0.70 y0.70 21.3Ž .9 Aharon Garden 4 y3.59 33.2 0.87 y0.62 17.3Ž .10 K.K.L. Street 5 y2.09 32.3 0.64 y0.15 7.2Ž .11 Herzl Street 11 y1.00 31.8 0.63 q0.75 y75.0

Ž .Average without streets 100 y2.74 32.7 0.653 y0.46 18.4

Table 7Ž . Ž .Cooling effect outside the site boundary 8C ; time: 1500 h averages for the days of measurement

Ž . Ž .Site Outside observation Orientation T 8C Cooling effects Kr

point Distances from site boundary

Border 20 m 40 m 60 m 80 m

Ž .1 Hayeled Avenue Hashtil St. E–W 31.8 2.3 2.0 1.3 0.8 0.2Ž .1 Hayeled Avenue Square S–N 31.8 1.9 1.8 1.6 1.2 0.5Ž .2 Meltz Garden Amsterdam St. E–W 33.2 0.6 0.6 0.6 0.5 0.4Ž .3 Emanuel Avenue Amsterdam St. E–W 33.2 2.7 0.9 0.5 0.3 0.3Ž .3 Emanuel Avenue Emanuel St. S–N 33.2 2.9 2.3 1.8 1.5 –Ž .4 Rothschild Avenue Bar Ilan St. E–W 32.3 1.9 1.8 1.0 0.7 –Ž .4 Rothschild Avenue Bar Ilan St. W–E 32.3 1.9 1.8 1.4 1.1 –Ž .5 Hen Avenue Hashoftim St. E–W 32.3 2.2 2.2 0.4 0.1 –Ž .5 Hen Avenue Hashoftim St. W–E 32.3 2.2 2.2 0.7 0.4 –Ž .6 Borochov Square K.K.L. st. S–N 32.3 2.9 2.1 1.1 0.1 0.1Average – – 32.5 2.15 1.77 1.04 0.67 0.30

w xFig. 7. Cooling effects outside the site versus distance from the site boundary K ; time: 1500 h.

( )L. Shashua-Bar, M.E. HoffmanrEnergy and Buildings 31 2000 221–235 233

Table 8Results of regression analysis of cooling effect outside the sites; time: 1500 h

Ž .Site Orientation n Buffer zone m

Ž .1 Hayeled Avenue E–W 30 y0.822 0.275 0.270 10.2Ž .1 Hayeled Avenue S–N 40 y0.681 0.179 0.172 10.4Ž .2 Meltz Garden E–W 42 y0.750 y0.015 0.073 y2.1Ž .3 Emanuel Avenue E–W 20 y0.902 y0.013 0.280 y0.5Ž .3 Emanuel Avenue S–N 32 y0.769 0 0.121 0Ž .4 Rothschild Avenue E–W 24 y0.786 0.113 0.188 6.0Ž .4 Rothschild Avenue W–E 20 y0.900 0.055 0.121 4.1Ž .5 Hen Avenue E–W 20 y0.906 0.384 0.524 7.3Ž .5 Hen Avenue W–E 20 y0.783 0.255 0.380 6.7Ž .6 Borochov Square S–N 30 y0.886 0.176 0.392 4.5Total 278

The average cooling effect at 1500 h outside the siteboundary is shown for each site in Table 7. Here thecooling effect at each observation point is the average forall days of measurement and is listed without the minussign. On the average, the cooling effect is 2.15 K at theboundary and drops to less than 0.5 K at a distance of 80m.

Fig. 7 shows in a graphic way the relationship betweenthe wooded site cooling effect on the surrounding area vs.the distance from the site boundary. The graph is drawn soas to pass through the averages at the bottom of Table 7and thus shows a general decay trend.

A decay function of the exponential type was proposedto fit the data in the following form:

DTŽ syr . ,Õsexp a yb s 8Ž . Ž .d d

DTŽoyr . ,Õ

where Õ is the site, subscripts o, s, r denote respectivelyŽ .the boundary point, the point at distance s in 10 m, units

from it, and the reference point, T is the air temperatureŽ .8C , DT sT yT , a is the constant term, and bŽ syr .Õ s,Õ r ,Õ d d

is the decay rate.Ž .The linearized form of Eq. 8 was estimated separately

for each site by the regression method. The regressioncoefficients are given in Table 8. All coefficients andcorrelations are statistically highly significant. The obser-vations used in these regressions are the daily coolingmeasurements at 1500 h.

The estimated decay rate b for the various sites is notuniform. This is as expected since the attenuation effectdepends on the size of the site, its orientation and itsgeometric configuration. The important fact, however, is

that for all these sites the cooling effect vanishes at about100 m from the boundary. Such being the case, theaverage b in these studied sites, is meaningful.

Ž .The linearized form of Eq. 8 , estimated by the regres-sion method, for all the sites together is:

DTŽ .syr , jln s0.1305y0.2313s 9aŽ .ž /DTŽ .oyr , j

or, equivalently:

DT s1.140DT 0.794 s 9bŽ . Ž .Ž syr . , j Žoyr . , j

Ž . Ž .where exp 0.1305 s1.140 and exp y0.2313 s0.794.The correlation coefficient rsy0.814 is highly signifi-cant for the 278 observations.

Ž Ž ..The decay rate is on the average 0.231 Eq. 9a , for aunit of 10 m, or b s0.0231 when the distance is ex-d

pressed in meters. Equivalently, the cooling effect dropssuccessively by a multiplier factor of 0.794 for every 10 mŽ Ž ..Eq. 9b . In other words, the cooling effect drops by

w Ž .x20.6% 100 1y0.794 every 10 m. The actual averagecooling effect is compared with that estimated by the

Ž Ž ..decay function Eq. 9b in Table 9. The effect is perceiv-able up to about 100 m from the edge of the site.

The last column in Table 8, the ‘‘buffer zone’’, showsthe distance from the site boundary at which the decayprocess of the cooling effect starts. This distance isŽ .a rb =10 m, and is arrived at by equating the right-d d

Žhand side of relation 8 to zero and solving for s in units of.10 m . On the average, the buffer distance is about 5.6 m

from the site border, with little variation among the sites.The largest buffer distance is found in the Hayeled Avenueand is 10 m.

Table 9Ž .Measured vs. estimated cooling effect with respect to distance from site boundary K ; time: 1500 h

Site boundary 20 m 40 m 60 m 80 m 100 m

Average measured values 2.15 1.77 1.04 0.67 0.30 –Ž Ž ..Estimated values by decay function Eq. 9a 2.15 1.54 0.97 0.62 0.38 0.24

( )L. Shashua-Bar, M.E. HoffmanrEnergy and Buildings 31 2000 221–235234

The influence of a large wooded site on the surroundingw xair temperature was found to depend on its size 4,6 . For

Žsmall green sites like those studied here width 20 to 60 m,.see Table 1 , the perceivable influenced distance is small,

about two to four times the width of the site. In largeŽ .wooded sites, such as Chapultepec Park 500 ha in Mex-

w xico City, Jauregui 5 found that the influence of this parkreaches a distance of about 2 km, about the same as thepark’s width. Assuming that the cooling influence of the

Ž .park follows a decay function as per Eq. 8 , the coolingŽ .effect will drop by a factor of 0.986 about 1.4% every 10

Ž .m equivalent to a factor of 0.75 every 200 m . Thus, if thecooling effect inside the park is say y3 K, then the effectat a distance of 1 km is small, about y0.7 K, but at adistance of 0.5 km it is about 1.5 K, which is quitesignificant. 7

6. Summary and conclusions

In this study, we investigated the cooling effect at 11small urban green sites with trees. The sites studied herehave various geometric configurations: two gardens, fouravenues, one green square, two courtyards and two streets.The analysis was carried out on measured air temperaturedata at noon gathered during July–August 1996, in theTel-Aviv urban complex.

The average air temperature at the surrounding areasoutside near the sites was 32.78C, at 1500 h, with relativelysmall deviation from site to site during the period ofmeasurement. The average cooling effect in all sites wasabout 2.8 K, ranging from as low as 1 K in a street with

Žheavy traffic to as high as 4 K in the smallest garden 0.15.ha . The width of the sites ranges from 20 to 60 m.An empirical model was developed for prediction of the

cooling effect inside the sites, based on the statisticalanalysis carried out on the 714 experimental observationsgathered each hour from the sites.

The following effects were found to be statisticallysignificant. They are of special interest for the design ofsmall urban green habitats.

Ø The background effect. The cooling effect in awooded site was found to depend, among other factors, onthe air temperature of its background outside the site. Thehigher this temperature, the stronger the cooling effect.Thus, we would expect a much stronger cooling effect ofabout 6 K in a typical garden in the southern part of IsraelŽ . 8say at Eilat , as against 2.8 K in the Tel-Aviv region.

The estimated relationship between the cooling effect ofa site and the background temperature, as proposed in this

7 Ž .In relation 8, we assume exp a s1, DT sy3 K, DT sy0.2 K,d o sŽ .200then y0.2sy3 R and solve for Rs0.986.

8 Eilat is about 108C hotter in summer than Tel-Aviv, and the relativehumidity about 20% at noon.

work, provides the means for proper comparison of siteswith different background temperatures. To our knowl-edge, the proposed technique is novel.

Ø The tree shading coÕerage. As expected from previ-ous studies in non-wooded urban spaces, the statisticalanalysis of our data indicates also that the shade factorplays a major role in determining the cooling effect of thesite. Its effect is more or less the same for all 11 sites. Inthe studied sites, shading in summer is provided by thetrees: on the average, about 80% of the cooling effect wascontributed by tree shading.

The shading coverage factor, besides its uncontestedrole in the cooling process, is also a control variable. It canbe regulated by the cultivation regime and by pruning, andin new sites by proper choice and placement of the shadetrees.

Following the empirical model, a more general analyti-cal model is being developed based on the cluster thermal

Ž .time constant CTTC model. This model will consider thetheory leading to the above two effects.

Ø The site specific effect. The site specific effect standsfor the effects of many unaccounted variables such as treecharacteristics, the site’s geometric configuration, the wa-ter regime, etc. A priori, we would have expected a majorrole for this factor. However, apart from the trees’ shadingeffect, what remains to be explained is minor. On theaverage, the specific effect contributes about 0.5 K ofcooling in addition to the shading effect. The variation ofthe specific effects among the sites is small.

Ø The effect of trees in the street. The shading effect oftrees in the streets was found to have the same magnitudeof cooling effect as in the other sites. However, heavy

w xtraffic has an opposite effect of about 2 K 13 . This canexplain the fact that the specific effect of Herzl Street is0.75 K compared to y0.5 K, the average of the sites. Thetwo sites Rothschild and Hen Avenues are also streets withheavy traffic, but with no noticeable heating effect. Thismay be due to the fact that these two avenues are wideŽ . Žabout 40 m and their tree canopies relatively high 10 to

.15 m high . The absence of heating effects in these twosites suggests that ventilation can be an important factorand should be taken into consideration in the design oftrees in a street. In any case, it is important to stress thefact, as found in this study, that even a moderate tree

Ž .shading coverage say 60% as in the Herzl Street morethan offsets the heating effect of heavy traffic.

Ø The cooling effect on the site surroundings. Therange of the cooling effect was found to be rather narrowand is perceivable up to 100 m from the site boundary.This fact corroborates earlier studies, although in our casewe are dealing, by design, with much smaller green areasnot wider than 60 m. The cooling values were found tofollow an exponential decay function.

The cooling effects of small green areas as found in thisstudy are significant. The effects are, however, local, andas such can be used to suggest some policy measures for

( )L. Shashua-Bar, M.E. HoffmanrEnergy and Buildings 31 2000 221–235 235

alleviating the so-called urban ‘‘heat island’’ effect in theurban environment.

Ž .a The range of the cooling effect being perceivable upto 100 m suggests small gardens, 200 m apart. Thesegardens can be designed to accommodate the recreationalneeds of young children and senior citizens. The proposedsize of such gardens is 0.1 ha, equal to the area of theapartment building commonly found in the urban Tel-Avivcomplex.

Ž .b The cooling effect of trees in streets was found to besignificant. In a street with trees, with heavy traffic such asin Herzl Street, the cooling effect reaches about 1 K.Streets make up more than 25% of the urban city area.This policy measure, properly designed, is most effectivein reducing the traffic heating effects. The cost is minimal.

Ž . 9c This study endorses Rosenfeld et al.’s suggestionfor at least one shade tree per eligible house to offset someof the cars’ parking effect in the courtyard.

Ž .d No consistent significant differences were foundbetween the sites vapour pressure and their respectivereference points. However, the fact that the leaves surfacetemperature, at noon, below and above the canopy waslower than the site reference point is due probably to windover the canopies and to evapotranspiration.

Acknowledgements

The authors are indebted to E. Goldberg for editionalassistance and comments and also to the two referees andto Prof. A.H. Rosenfeld who provided many useful sugges-tions.

9 w xRosenfeld et al. 2 suggest three trees per eligible house.

References

w x1 I. Saito, O. Ishihara, T. Katayama, Study of the effect of green areason the thermal environment in an urban area, Journal of Energy and

Ž .Buildings 15–16 1990 443–446.w x2 A.H. Rosenfeld, H. Akbari, S. Bretz, B.L. Fishman, D.M. Kurn, D.

Sailor, H. Taha, Mitigation of urban heat islands: materials, utilityŽ .programs, updates, Journal of Energy and Buildings 22 1995

255–265.w x3 A.H. Rosenfeld, G. Romm, H. Akbari, M. Pomerantz, Coom com-

munities strategies for heat island mitigation and smog reduction,Ž .Journal of Energy and Buildings 28 1998 51.

w x4 F. Wilmers, Effects of vegetation on urban climate and buildings,Ž .Journal of Energy and Buildings 15–16 1988 507–514.

w x5 E. Jauregui, Influence of a large urban park on temperature andconvective precipitation in a tropical city, Journal of Energy and

Ž .Buildings 15–16 1990 457–463.w x6 V.T. Ca, T. Asaeda, E.M. Abu, Reduction in air conditioning energy

Ž .caused by a nearby park, Journal of Energy and Buildings 29 199883–92.

w x7 B. Givoni, Comparing temperature and humidity conditions in anurban garden and in its surrounding areas, Interim Report no. 2,Building Research Station, Technion, 1972.

w x8 T. Honjo, T. Takakura, Simulation of thermal effects of urban greenareas on their surrounding areas, Journal of Energy and Buildings

Ž .15–16 1990 443–446.w x9 N. Sharlin, M.E. Hoffman, The urban complex as a factor in the air

temperature pattern in a mediterranean coastal region, Journal ofŽ .Energy and Buildings 7 1984 149–158.

w x10 T.R. Oke, The micrometeorology of the urban forest, J. Phil. R. Sec.Ž .Land. B 324 1989 335–349.

w x11 Y. Nakamura, T.R. Oke, Wind, temperature and stability conditionsin an east–west oriented urban canyon, Journal of American Envi-

Ž .ronment 22 1988 2691–2700.w x12 H. Swaid, M.E. Hoffman, Climatic impacts of urban design features

for high and mid-latitude cities, Journal of Energy and Buildings 14Ž .1990 325–336.

w x13 H. Swaid, Urban climate effects of artificial heat sources and groundshadowing by buildings, International Journal of Climatology 13Ž .1993 797–812.